Redacted for Privacy

Transcription

1 AN ABSTRACT OF THE THESIS OF Karen M. Theiling for the degree of Master of Science in Entomology presented on April 10, Title: The SELCTV Database: The Susceptibility of Arthropod Natural Enemies of Agricultural Pests to Pesticides. Abstract approved: Redacted for Privacy /- Documentation of the side effects of pesticides on arthropod natural enemies has expanded rapidly since the 1950's as part of an increase in non-target side effects literature. Most reviews have been based on empirical analysis of selected literature. The SELCTV database was developed to make a larger information base accessible for characterization and analysis. The feasibility of such a database is a function of improving microcomputer technology and database management software. Record structure and scope of the SELCTV database included 40 information fields covering natural enemy biology, pesticide chemistry, toxicology and literature citations. SELCTV was assembled from over 900 published papers, believed to constitute 80-90% of available literature through the early 1980's. Currently, some 12,600 records contain taxonomic, biological, toxicological, reference and summary information for over 600 species of natural enemies in 88 families. Research was conducted in 58 countries around the world and included predators and parasitoids associated with 60 agricultural commodities. All major classes of pesticides are represented, including microbial insecticides. The impact of over 400 agricultural chemicals on natural enemies by means of one of ten basic test types has been distilled into SELCTV.

2 Many different types of natural enemy responses were reported in the literature. In addition to recording these as documented, measurements were translated to a scale ranging from 1 (0% effect) to 5 (90-100% effect). This toxicity rating scale formed the basis for most SELCTV analysis. Selectivity ratios, resistance ratios and sublethal effects were other types of data which were recorded when possible. Lethal and sublethal effects were evaluated for many species, pesticide and test method data groupings. Results showed that predators were less susceptible and more variable in responses to pesticides than parasitoids. Relative susceptibility was computed for important natural enemy species. Among the most tolerant were Lycosa pseudoannulata, Cryptolaemus montrouzieri and Chrysopa carnea. Insecticides were the most toxic of pesticide classes, followed by herbicides, acaricides and fungicides, respectively. Among insecticide classes, a trend of increasing toxicity to natural enemies was demonstrated from the early inorganics to synthetic pyrethroids. More recent microbials and IGR's were less toxic and more selective. In addition to characterizing the natural enemypesticide impact literature and conducting selected analyses, several case studies were compiled to demonstrate application of SELCTV to decision making in pest management. Another compares results of SELCTV with a large standardization testing program from Europe. Increased size and degree of specificity of the information base were among research trends elucidated through chronological searches of SELCTV. Specific natural enemies, pesticides and test methods as assessment components have fluctuated relative to pesticide use, as well as testing and pest management philosophies. The study of diverse natural enemy responses to pesticides has

3 led to the identification of unique factors that influence natural enemies in different ways or to a greater or lesser degree than pests. Differences in the susceptibility of pests, predators and parasitoids are discussed.

4 The SELCTV Database: The Susceptibility of Arthropod Natural Enemies to Agricultural Pests to Pesticides by Karen M. Theiling A THESIS submitted to Oregon State University in partial fulfillment of the requirements for the degree of Master of Science Completed April 10, 1987 Commencement June 1987

5 APPROVED: Redacted for Privacy t)rofessor o /// Entomology in charge of major Redacted for Privacy Head/of Department)of Entomology Redacted for Privacy tdean of Grad e School Date thesis is presented April 10, 1987

6 TABLE OF CONTENTS INTRODUCTION 1 MATERIALS and METHODS 6 Hardware/Software 6 Record Structure 6 Inputting Data 10 Characterization and Analysis 11 RESULTS 15 Characterization of the database 15 By reference attributes 15 By natural enemy attributes 21 By pesticide attributes 35 By test and summary attributes 40 Analysis of the database 56 Toxicity rating analysis 56 By natural enemy attributes 56 By test attributes 68 By pesticide attributes 68 Analysis on other fields 81 Case studies 92 Average toxicity of chemical groups for selected crops 92 Cotton selectivity table 96 Synthetic pyrethroid toxicity to cotton and apple natural enemies 99 Comparison of susceptibility assessments: IOBC and SELCTV 104 DISCUSSION 107 Literature documenting natural enemy susceptiblity to pesticides 107 Database management as a means of census and analysis 109 Trends and new research questions 111 Natural enemy-pest representation 111 Pesticide representation 114 Pesticide toxicity and selectivity 115 Patterns in pesticide susceptibility 117 Testing and research: patterns and considerations 120 CONCLUSIONS 124 BIBLIOGRAPHY 127 APPENDIX A 141

7 LIST OF FIGURES Figure Page 1. Incidence of records of pesticide impact on arthropod natural enemies over time from the SELCTV database Incidence of publications documenting pesticide impact on arthropod natural enemies over time from the SELCTV database Distribution of SELCTV database records of pesticide impact on arthropod natural enemies for the 20 most commonly tested families Distribution of SELCTV database records of pesticide impact on arthropod natural enemies grouped by life stage Distribution of SELCTV database records of pesticide impact on arthropod natural enemies for the 16 most common crop categories Distribution of SELCTV database records of pesticide impact on arthropod natural enemies grouped by research location Incidence over time of pesticide impact research on arthropod natural enemies for each major pesticide class from the SELCTV database Incidence over time of insecticide impact research on arthropod natural enemies for major insecticide groupings from the SELCTV database Distribution of SELCTV database records of pesticide impact on arthropod natural enemies by pesticide exposure time Distribution of SELCTV database records of pesticide impact on arthropod natural enemies grouped by test method Changes in precision associated with pesticide impact assessment on arthropod natural enemies measured in records of field vs. median lethal assays over time from the SELCTV database Incidence of predominant test methods used to assess natural enemy susceptibility to pesticides in number of records over time from the SELCTV database 46

8 13 Distribution of SELCTV database records of pesticide impact on arthropod natural enemies grouped by evaluation time period Distribution of SELCTV database records of pesticide impact on arthropod natural enemies grouped by toxicity rating Distribution of SELCTV database records of pesticide impact on arthropod natural enemies grouped by data rating Incidence over time of documented sublethal side effects of pesticides on arthropod natural enemies from the SELCTV database Mean insecticide toxicity ratings versus variance for 12 families of parasitoids in the SELCTV database Mean insecticide toxicity ratings versus variance for 28 families of predators in the SELCTV database Average toxicity ratings and 95% confidence intervals for insecticide chemical groups tested for side effects on arthropod natural enemies from the SELCTV database Average toxicity ratings and 95% confidence intervals for acaricide chemical groups tested for side effects on arthropod natural enemies from the SELCTV database Average toxicity ratings and 95% confidence intervals for fungicide chemical groups tested for side effects on arthropod natural enemies from the SELCTV database Average toxicity ratings and 95% confidence intervals for herbicide chemical groups tested for side effects on arthropod natural enemies from the SELCTV database Distribution of toxicity ratings among insecticide chemical groups assessed for side effects on arthropod natural enemies from the SELCTV database Distribution of toxicity ratings among microbial/biotic insecticides assessed for side effects on arthropod natural enemies from the SELCTV database 76

9 25. Distribution of toxicity ratings among carbamate pesticides assessed for side effects on arthropod natural enemies from the SELCTV database Average toxicity ratings for insecticide chemical groups to arthropod natural enemies of apple and cotton from SELCTV database records Average toxicity ratings for synthetic pyrethroids across all natural enemy species and crops in the SELCTV database Average toxicity ratings for synthetic pyrethroids to arthropod natural enemies of apple and cotton from the SELCTV database 103

10 LIST OF TABLES Table Page 1. Record structure and field descriptions for SELCTV and REFERENCE databases of documented pesticide impact on arthropod natural enemies 7 2. Frequency of keywords from the SELCTV database describing aspects of arthropod natural enemy susceptibility or resistance to pesticides Distribution of SELCTV database records of pesticide impact on arthropod natural enemies grouped by order Distribution of SELCTV database records of pesticide impact on arthropod natural enemies for the 20 most commonly tested genera Distribution of SELCTV database records of pesticide impact on arthropod natural enemies for the 22 most commonly tested species Three natural enemy genera most commonly tested for pesticide side effects on each of the 10 most prevalent crops from the SELCTV database Relative distribution of SELCTV database records by host or prey associated with arthropod natural enemies tested for pesticide impact Frequency and classification of the 30 most commonly studied pesticides in the SELCTV database of pesticide impact on arthropod natural enemies Relative distribution of SELCTV database records of pesticide impact on arthropod natural enemies grouped by pesticide formulation type Distribution of SELCTV database records of pesticide impact on arthropod natural enemies sorted by pesticide chemical group Documented sublethal pesticide side effects and their distribution among natural enemy orders from the SELCTV database Frequency of selectivity ratios and assigned selectivity ratio index values indicating comparative pest/natural enemy pesticide susceptibility from the SELCTV database 55

11 13. Average toxicity ratings and variances of pesticide classes to all arthropod natural enemies, predators and parasitoids from the SELCTV database Average toxicity ratings of pesticide classes to the 15 most commonly tested natural enemy families from the SELCTV database Average susceptibility of the 22 most commonly tested natural enemy species to all pesticides from the SELCTV database Average susceptibility and variance for all arthropod natural enemies, predators and parasitoids grouped by life stage from the SELCTV database Average toxicity ratings and variance for test methods used to assess pesticide impact on natural enemies from the SELCTV database Frequency and average toxicity ratings of common pesticides tested for side effects on arthropod predators and parasitoids from the SELCTV database Average toxicity ratings and pesticide classification for the most toxic pesticides to arthropod natural enemies from susceptibility assessments in the SELCTV database Frequency and average toxicity ratings for the least toxic insecticides to arthropod natural enemies in the SELCTV database Frequency and average toxicity ratings for the least toxic acaricides to arthropod natural enemies in the SELCTV database Frequency and average toxicity ratings for the least toxic fungicides to arthropod natural enemies in the SELCTV database Average pesticide resistance ratios for arthropod natural enemies from the SELCTV database grouped by family Average pesticide resistance ratios for arthropod natural enemies from the SELCTV database by pesticide chemical group and compound 85

12 25 Average selectivity ratios by natural enemy order for arthropod pests and natural enemies from comparative assessments of pesticide susceptibility in the SELCTV database Average selectivity ratios by natural enemy family for arthropod pests and natural enemies from comparative assessments of pesticide susceptibility in the SELCTV database Average selectivity ratios by natural enemy and pesticide groupings for arthropod pests and natural enemies from comparative assessments of pesticide susceptibility in the SELCTV database Average selectivity ratios by compound for arthropod pests and natural enemies from comparative assessments of susceptibility in the SELCTV database Average magnitude of sublethal pesticide side effects individually and by natural enemy order from the SELCTV database Average susceptibility of natural enemies from four cropping systems to insecticide groups from SELCTV database records Average toxicity of insecticides to natural enemy genera associated with cotton in the United States from SELCTV database records A comparison of susceptibility assessments from WPRS/IOBC standardized testing and SELCTV database records for common natural enemy genera-insecticide combinations 105

13 THE SELCTV DATABASE: THE SUSCEPTIBILITY OF ARTHROPOD NATURAL ENEMIES OF AGRICULTURAL PESTS TO PESTICIDES INTRODUCTION Since the advent of agricultural pesticides in the late 19th century, their impact on arthropod natural enemies of pests has only slowly been recognized (Debach 1964). Today, few are aware of the high levels of biological control that go on unaided until natural enemies are affected by chemical pesticides. Documentation of the deleterious effects of pesticides on arthropod predators and parasitoids, which first appeared in the early 1900's, has grown into a global literature base of at least 2000 publications (Croft, personal communication). Studies of pesticide side effects on natural enemies have evolved slowly and sometimes tangentially to similar studies with pest arthropods (Croft 1987). Before DDT, pesticides were commonly less effective, of lower potency and less widely used in US agriculture (Perkins 1982). Unilateral dependence upon them was the exception rather than the rule. Early pesticides were inorganic, heavy metals or derivatives of natural toxins (Van den Bosch and Stern 1962). Many acted as stomach poisons, primarily affecting plant feeding species; this conferred some behavioral or ecological selectivity to natural enemies (Ripper 1956). Synthetic organic pesticides radically improved crop protection, allowing yields not previously possible. However, this efficacy was not without greater negative contingencies to biological control (Stern et al. 1959).

14 Early studies showed that the composition of arthropod community structure was altered by extensive synthetic pesticide use (Pimentel 1961). Natural enemy populations were temporarily decimated and highly susceptible species sometimes disappeared (Lord 1949, Ripper 1956, MacPhee and Sanford 1961). Following pesticide applications, natural enemy re-establishment lagged behind pests. This was due partly to consistent availability of pest food supplies compared to natural enemies, whose dependence on pest populations for food limited their recolonization (Bartlett 1964). Selection for pesticide resistance was accelerated by heavy reliance on pesticides (Ripper 1956). Still, the broad spectrum efficacy of synthetic organic pesticides was unparalleled. intensive pesticide treatment continued. Expansion of acreage under Over time, the rate and frequence of application increased as pests were released from their natural controls (Metcalf and Luckmann 1982). Pest resistance, resurgence and secondary outbreaks caused by synthetic pesticide use began to rise toward the end of the 1950's (Van den Bosch and Stern 1962). Integration of chemical and biological control became necessary when some pest populations could no longer be chemically limited (Ripper 1956). As a result, integrated pest management (IPM) was promulgated as a crop protection philosophy (Stern et al. 1959). Implementation of IPM was limited, in part, by the lack of selective chemical pesticides (Newsom et al. 1976, Gruys 1980, Mullin and Croft 1985). Additionally, the rate at which new broad spectrum compounds became available did not encourage conservative pesticide use. Overuse and eventual pesticide failure became sufficiently widespread to generate new scientific studies. Questions arose regarding the intrinsic susceptibility of natural enemies relative to pests. How might we spare natural enemies

15 while killing pests? Studies undertaken to assess the acute and chronic affects of pesticides on natural enemies began to increase in the early 1960's, in an upward trend that continues today (Croft 1977). Early reviews of pesticide impact on natural enemies featured toxicity summaries, identified deleterious ecological effects and discussed ways of achieving ecological selectivity (Ripper 1956, van den Bosch and Stern 1962, Bartlett 1964). Several comprehensive reviews of pesticide impact on arthropod natural enemies were published in the mid 1970's (e.g., Kirknel 1974), most recently by Croft and Brown (1975). The breadth of their review reflected the evolving status and specialization of pesticide impact studies on natural enemies in the 1960's and 1970's. Their discussion spanned from direct toxic effects of pesticides on natural enemies to indirect effects, such as sublethal and ecological influences, to pesticide resistance in natural enemies. Selectivity needs for pest management were outlined with newer pesticides including microbial insecticides and insect growth regulators. They concluded that improved selectivity with these agents was likely. The shortage of physiologically selective pesticides plagued IPM in the 1970's. Ecological selectivity remained the major means of preferentially sparing natural enemies, as it is today (Newsom et al. 1976, Hull and Beers 1985). Newsom's review of selectivity is one of the first to focus on more specialized aspects of the overall topic of natural enemy response to pesticides. Contemporary reviews have largely followed this trend, summarizing natural enemy response to pesticides: regionally (e.g., in the USSR, Kurdyukov 1980), by commodity (apple, Croft and Whalon 1982, Niemczyk et al. 1981; cotton, Sukhoruchenko et al. 1977), or by individual pesticide groups (e.g., synthetic pyrethroid insecticides,

16 Croft and Whalon 1982, Niemczyk et al. 1981; microbial pesticides, Flexner et al. 1986; insect hormone analogues, Beckage 1985). Other reviews have focused on standardized side effects testing (von Franz 1974, Croft 1977, Bogenschutz 1979, Franz et al. 1980, Hassan et al. 1983), modes of pesticide uptake (Croft 1977), physiological and ecological selectivity (Newsom et al. 1976, Gruys 1980, Hull and Beers 1985, Mullin and Croft 1985), resistance development (Croft and Strickler 1983, Fournier et al. 1986), genetic improvement of resistant strains (Roush 1979, Hoy 1985) and their use in biological control (Croft and Hoyt 1983, Hoy 1984). The literature base on natural enemy/pesticide interactions has expanded rapidly (Croft and Brown 1975, Croft 1977), having approximately doubled in the last decade (see Results). Evaluations have become more specific and methods more precise over this period (Croft 1987). As previously demonstrated, reviews published since the 1970's tend to focus on subdivisions within the subject area such as susceptibility testing, sublethal effects or pesticide selectivity. Because the literature of pesticide effects on natural enemies has become so vast, comprehensive summation by empirical or traditional methods alone has become less feasible. Modern microcomputer technology and database management software makes possible the development of searchable databases from large information sets. In this thesis, data published over the past 40 years documenting pesticide side effects on natural enemies of agricultural pests are summarized in the form of a large database (SELCTV). This database provides a means to count, index, sort and search for key assessments made on a wide variety of natural enemies (609 species). Currently, the total literature includes 933 published papers from 58 countries. SELCTV contains toxicity assessment data for

17 over 400 agricultural chemicals, including fungicides, herbicides, insecticides, acaricides, feeding repellents, insect and plant growth regulators and others. The principal objectives of this thesis were 1) to develop and assemble a database on the impact of pesticides on arthropod natural enemies of agricultural pests, 2) to characterize database contents by many attributes, including features of the natural enemy, the pesticide, the assessment method and test results, and 3) to use SELCTV to address scientific questions or hypotheses regarding the overall topic. For example, toxicity assessment summaries from SELCTV were developed using a variety of pest/crop/pesticide combinations. Selectivity tables for natural enemies in several cropping systems were assembled and a number of case history searches of SELCTV were performed to demonstrate uses of the database in problem solving. It was anticipated that construction of the database would provide an improved means for summarizing information on the effects of pesticides on arthropod predators and parasitoids. In addition, a structure and method would be provided for dealing with future publications and reviews of this topic.

18 Hardware/software MATERIALS AND METHODS The SELCTV database was originally developed on a Digital Rainbow 100 microcomputer (CP/M operating system) and Winchester 10 megabyte hard disk using dbase 11/86 version 2.4 (Ashton-Tate 1982), a relational database management system. Because of space and speed limitations, the database was transferred to an.at&t 6300 microcomputer (MS-DOS operating system) with a 20 megabyte hard disk. Software was upgraded to dbase III PLUS, version 1.0 (Ashton-Tate 1985). At this time reference citations were moved and expanded in a separate database called REFER. SELCTV and REFER are related on a key field, REFNUM (reference number), and are henceforth referred to collectively as SELCTV. A menu driven, compiled version of SELCTV is currently under development to circumvent the need for dbase III PLUS literacy to query, append to or edit SELCTV (for further information on this program, contact B. A. Croft). Record Structure SELCTV and REFER record structures (Table 1) were determined by consideration of attributes of the natural enemy tested, the pesticide used, testing methods employed and the literature reference for each data entry. These requirements were balanced against the consistent availability of data elements from one literature publication to the next. A total of 40 fields were selected to accomodate data, comprising the structure or template for each record. Four distinct groupings of database fields emerged and were organized accordingly in the record structure: 1) biological, 2) chemical, 3) test and summary, 4) reference citation (see Table 1 for detailed explanation of the contents for each field). Field

19 Table 1. Record structure and field descriptions for SELCTV and REFERENCE databases of documented pesticide impact on arthropod natural enemies. SELCTV Subdivision: Biological GENUS: SPECIES: FAMILY: taxonomic genus of the natural enemy being tested, e.g., Chrysopa. taxonomic species of the natural enemy being tested, e.g., carnea. taxonomic family of the natural enemy being tested, e.g., Chrysopidae. NE:ATTRIBU: coded field of 5 biological attributes of the tested natural enemy: 1) ORDER: taxonomic order of the natural enemy. 2) NE:TYPE: predator (d), parasitoid (r) or both (b). 3) SEX: female (f), male (m) or unspecified (u). 4) SOURCE: where the test natural enemies were obtained, (1) ab reared or (f) ield collected. 5) SUSCEPTIBILITY STATUS: resistant (r), tolerant (t), or neither (n). STAGE: lifestage of the natural enemy at the time of pesticide impact evaluation, e.g., egg(e), larva(1). PROD:UNIT: crop or production unit with which natural enemy is commonly associated, e.g., corn (cn), alfalfa (al), bean (bn), etc. LOCI: location where the research was conducted, e.g., USA-OR, Australia, Canada-BC, etc. HOST:PREY: HP:TOX:DAT: scientific or common name of host/prey of the tested natural enemy, e.g., Myzus persicae. was toxicity for the host or prey included in the publication? yes(y), no(n). SELCTV Subdivision: Chemical CPD:NAME: experimental or common chemical name of the toxicant tested, e.g., carbaryl, demeton.

20 Table 1 Continued FORMULATN: CHEM:CLASS: CHEM:GROUP: CHEM:RATE: CR:UNITS: CR:VALUE: concentration of active ingredient and/or type of preparation for use, e.g., 25%wp (wettable powder), 50%ec (emulsifiable concentrate). pesticide class to which the compound belongs belongs, e.g., fungicide(f), insecticide(i). chemical catagory by primary functional group or structure, e.g., organochlorine (oc), synthetic pyrethroid (sp). manner of application, e.g., dose (d), field(f), concentration (c), residue (r). units by which application was measured, e.g. grams ai/ac, lb/100liter, etc. numeric value associated with rate units. SC:AI: standard concentration in tai. SELCTV Subdivision: Test and Summary DUR:EXP: TST:METHD: EVAL:TIME: RESP:TYPE: amount of time natural enemy is in actual contact with the toxicant. method used for susceptibility assessment, e.g., contact with fresh, dry film (c), dip (d), field (f). time elapsed from initial pesticide exposure to assessment of impact, e.g., 48h (hours). response that is being measured, e.g., LC, LD, mortality, longevity, etc. RESP:UNITS: RESP:VALUE: response units, e.g., %+, %-, or units associated with LD, LC, LT assays. numerical value associated with RESP:UNITS. RR:RATIO: LD or LC 50 of resistant strain divided by tat of the susceptible strain. TOX:RATING: common scale of ranking natural enemy response, ranging from 0 (no effect) to 5 (90-100% mort).

21 Table 1 Continued DAT:RATING: SUBLETHAL: a two part index; the first digit is a relative indicator of precision, ranging from 1 (best) to 4 (worst). The second digit indicates presence (1) or absence (2) of statistics to support inference. effects of non-lethal doses, e.g. fecundity (fec) or longevity (long), rated on the same scale as toxicity. SLECTRATIO: LC 50 or LC50 of host or prey divided by the LD or LD 50 of associated natural enemy 50. (vgiues < T indicate selective compounds). COMMENTS: any other relevant features of the paper that are that are not included in database fields. REFERENCE Subdivision: Reference Citation REF:NUM: AUTHOR: PUB:DATE: unique number assigned to each reference, which allows citations to be related to appropriate records in SELCTV. author(s) of publication from which the data is extracted. year of publication. TITLE: JOURNAL: VOLUME: PAGES: title of the publication. scientific or technical journal of pubication. volume and number of journal. page numbers of publication. LANGUAGE: ABSTRACT: RECORDS: KEYWORDS: language paper was published in. coded field indicating type of literature, e.g. abstract only, or paper and abstract, etc. number of records extracted from publication. abbreviations which correspond to subjects of interest in natural enemy susceptibility assessment, e.g. physiological selectivity p- sel), etc.

22 10 lengths were set by anticipating needs for space based on possible field contents. Data type designations included character, numeric and logical, as allocated in dbase. Data type was dictated in part by the degree of specificity needed in the field and by whether or not numerical computations were to be made on the information in a particular field. Nested groupings of fields were created for natural enemy and pesticide information to build in levels of search specificity. included ORDER/ FAMILY/ GENUS /SPECIES fields and These combinations CHEM:CLASS (pesticide class)/ CHEM:GROUP (chemical group)/ CPD:NAME (compound name) fields, respectively. Inputting Data Publications entered in SELCTV were obtained from B. A. Croft, who maintains an extensive collection of literature pertaining to the side effects of pesticides on arthropod natural enemies. This literature base was used previously for the review published by Croft and Brown (1975). It was initiated and has been updated by means of computerized library searches of abstract databases such as BIOSIS, CAB and Agricola, and by cross checks of bibliographies from acquired publications. In addition, periodic searches of the Review of Applied Entomology have produced relevant literature. Collected literature is believed to be essentially complete for the United States and most western countries. However, SELCTV is limited with respect to eastern European, Russian and far eastern literature. This information was generally less available due to language barriers and literature inaccessibility. However, abstracts have been obtained for much of this literature and data gleaned from them as possible. Data were input into SELCTV by extracting relevant information from a publication and filling each record as completely as possible. A new record was appended for

23 11 every unique documentation of a natural enemy's response to any pesticide and formulation at a given dose on a particular crop using a specific test method, exposure time and evaluation time. The number of records extracted from a publication was highly variable, ranging from 1 to 600 (for tabulations of toxicity from reviews). Because the form of information in the literature did not always coincide with SELCTV structure, an element of interpretation was occasionally introduced. While unavoidable, this subjectivity is believed to be relatively minor over the entire database. Care was taken to preserve data integrity and to maintain consistency within SELCTV. To date, 933 publications have been distilled into the database and SELCTV will be expanded as data become available. Characterization and Analysis After completion of data entry, the contents of each field were screened to correct errors and insure consistency. Fields were then characterized. This involved programming in dbase for required counts or listings, usually on indexed fields. Fields with a limited number of possible entries were characterized using simple counting routines, while indexing and sorting field contents were used to simplify summation on fields with many possible entries. Programs were then designed to list unique field entries and their frequencies. All fields were similarly characterized. More complex characterization was then undertaken by creating multiple field indexes and more complex programs. For example, the frequency of natural enemy genera by crop was examined by indexing SELCTV on PROD:UNIT (production unit) as a primary key and GENUS as a secondary key. A program was then written to list and count data by crop, and within each crop by genus of natural enemy.

24 12 Publication date was used to identify trends in research priorities and response data over time. Field contents were partitioned over 5 year increments and results displayed using line charts. Most SELCTV analyses involved calculations and comparisons of average toxicity ratings (TOX:RATING) and associated variance components for data grouped by selected criteria. These calculations provided a means to examine factors which affect natural enemy responses. Data were ordered by building indexes on single or multiple keys. Average toxicity and variance values were then calculated using programs written in dbase. Additional analyses were performed on other numeric indices within the database, such as resistance ratios (RR:RATIO), selectivity ratios (SLECTRATIO) and sublethal effects ratings. Computations were then performed on data grouped by various criteria such as pesticide or natural enemy attributes. The collected literature of pesticide impact on arthropod natural enemies is sufficiently complete to be considered a population (see above and later discussion of how comprehensive SELCTV might be with respect to the global literature). On the basis of this assumption, the use of statistical methods for inference was not appropriate. Rather, descriptive statistics such as histograms or scatterplots have been employed to represent data. All response data were translated to a unitless scale (TOX:RATING) ranging from 1 to 5, which allowed a common rating of severity to be imposed upon many different types of natural enemy responses to pesticides. Toxicity ratings on a scale of percent mortality or effect were not linearly incremented. Extremes in effects were considered most important to document as these indicate cases of preferential selectivity or of high hazard to natural

25 13 enemies. Hence, ratings were more heavily assigned at scale extremes. While the range of possible analyses in SELCTV was broadened considerably by using a common scale, computed variance values were altered. These computed variance values represent a relative approximation of variability about the true mean because SELCTV calculations were performed on a different scale (TOX:RATING) than the original response measurement. High variance indicated a greater frequency of extreme values on the toxicity scale, whereas means alone gave no indication of the spread of data points. In some cases, variance was not biologically meaningful. In part, variance reflects the number of different investigators, test methods, natural enemies, pesticides and doses present in the mean toxicity value computed from a data grouping. Additionally, where toxicity means are high on the scale (4-5), the range of variability is limited by the upper end of the toxicity scale. Variance may be underestimated in these cases, principally reflecting variation below the mean. Confidence associated with SELCTV analyses is primarily dependent upon: 1) the size (n) of specific data groupings, 2) numerical differences between means based on the specificity of the search or computation and 3) the number of fields or data elements spanned by an average value. There also may be hidden data trends or research biases which influence mean values. These components are identified and their impact estimated in the discussion section. The biological relevance of means and variances in SELCTV is relative at a broad level of analysis. Here, SELCTV indicates trends in natural enemy research, biology or ecology. Trends or differences which persist across several to many parameters (e.g., species and/or pesticide and/or test method) may be biologically significant, and

26 14 merit further experimental investigation. As search criteria become more general, the number of parameters incorporated in the mean increases as does the number (n) of data points which satisfy these more general search criteria. Extremes are assumed to balance each other at a broad level of generalization. Higher variance expected for means of broad data groupings is offset somewhat by a larger number of data points (n). With specific comparisons or calculations, results can be interpretted more literally. The number of records which satisfy these criteria and the number of parameters spanned by means become small. These mean values have more direct and precise application to the field or testing conditions specified. However, means of small data subsets or groupings are strongly influenced by results from individual studies. Caution must be used in interpreting analysis based on small data sets. Variance of small data sets is a good indicator of the consistency or range spanned by data points.

27 15 RESULTS Contents of the SELCTV database were initially characterized for each of the 40 fields of the record structure (Table 1). This included the current 12,593 records in SELCTV and the related 933 records in REFERENCE. Following a discussion of PUB:DATE, AUTHOR, JOURNAL, LANGUAGE and KEYWORDS fields, the remaining fields were treated in the order of their appearance (Table 1). Because the study of pesticide impact on natural enemies is an evolving field, time trend curves were used to show how documentation of different aspects has progressed. Analysis of the database was based on numerical computations on the TOX:RATING (Toxicity rating), RR:RATIO (Resistance ratio), SLECTRATIO (Selectivity ratio) and SUBLETHAL (Sublethal effects) fields. Analysis follows characterization of SELCTV in the Results section of this thesis. Finally, four historical cases of analysis were prepared to illutrate uses of SELCTV. These included a comparison of the susceptibility of natural enemies from 4 crops to insecticide chemical groups (CHEM:GROUP), a selectivity table for 8 important genera of natural enemies found on cotton and a case study of the impact of synthetic pyrethroids on apple and cotton natural enemies. Finally, SELCTV results have been compared with those compiled from a major European study of pesticide impact on arthropod natural enemies, which featured the development of standardized assessment techniques. Characterization of the Database By Reference Attributes The distribution of published research over time (PUB:DATE) was examined from several perspectives. Initially, frequency histograms were constructed for the

28 16 number of records and number of publications on a yearly basis. Both varied considerably from year to year. Therefore, representations were grouped into classes of 5 year periods were plotted (see Fig. 1 and 2). The number of records per period in SELCTV (Fig. 1) began to rise after 1945, with a sharp upturn in slope after This rate of increasing research has been maintained thereafter, with the exception of A decrease in records per year occurred in the period. This was believed to be an artifact of incomplete literature acquisition for the recent past, and applies to all time trend figures (see later discussion). The maximum records generated during any one year was 896 in The distribution of publications over time (Fig. 2) revealed a similar trend with publication rate as seen with records. Three distinct rate phases were distinguished. In the first, lasting from 1940 to 1955, the number of publications increased at a barely discernible rate. From 1955 to 1970, the publication rate increased moderately and steadily. After 1970, there was a near exponential increase in the number of pesticide impact studies published for arthropod natural enemies. Again, the decrease in publication rate after 1980 was attributed to ongoing literature acquisition for the most recent past. The maximum number of publications in SELCTV for a given year was 77 in Records for chemical pesticides and microbial pesticides were maintained separately in SELCTV. Thirtytwo publications contained both chemical and microbial pesticide assessments. Hence, while there were 843 chemical pesticide and 123 microbial pesticide citations, the number of unique publications in SELCTV was 933 (Complete literature citations in appendix A). Most

29 cn 2400 C3 X C) Li IT mx = FIVE YEAR INTERVAL ENDING... Figure 1. Incidence of records of pesticide impact on arthropod natural enemies over time from the SELCTV database.

30 Cl) z I--.c - u I 200 m =m u- 160 o m w co x 120 z FIVE YEAR INTERVAL ENDING... I I I I Figure 2. Incidence of publications documenting pesticide impact on arthropod natural enemies over time from the SELCTV database.

31 19 prolific first authors by number of publications were B. A. Croft (12 publications), M. A. Hoy (10), S. A. Hassan (9), H. K. Kaya (8) and E. Niemczyk and S. K. Wiackowski with 6 each. First authors by number of records in SELCTV included K. H. Sanford (625 records), B. R. Bartlett (535), S. A. Hassan (406), A. W. MacPhee (339) and W. E. Ripper (282). The latter group of authors published reviews of pesticide impact on natural enemies (Ripper 1956, MacPhee and Sanford 1961, Bartlett 1964, Hassan 1982, Hassan et al. 1983). These included extensive toxicity summaries which generated many records. Papers containing data input into SELCTV were cited from 275 scientific journals, reviews, technical publications and congress or symposium proceedings published in 19 languages. Journals most common in bibliographic citations included: J. Econ. Entomol. (156 citations), Environ. Entomol. (97), Can. Entomol. (30), Z. Angew. Entomol. (30) and Entomophaga (24). Literature was most often published in English, followed by German, Russian and French, respectively. Twelve percent of records were extracted from abstracting journals or English summaries of papers in foreign languages. Keywords (KEYWORDS) were assigned to each publication as applicable by subject area. The most frequently assigned keyword was that for physiological selectivity of pesticides to natural enemies (Table 2). Physiological selectivity was discussed in many publications as a desirable pesticide feature for use in IPM. However, documentation of physiological selectivity to natural enemies has been far less common than implied by KEYWORDS counts. Next, in decreasing order of frequency were keywords indicating factors influencing susceptibility, mode of pesticide uptake, methods development for susceptibility assessment, general pesticide selectivity, ecological selectivity and resistance documentation.

32 20 Table 2. Frequency of keywords from the SELCTV database describing aspects of arthropod natural enemy susceptibility or resistance to pesticides. Symbol Topic No.Records p-sel Physiological selectivity 7081 fi Factors affecting susceptibility (S) 2360 md-s Methods development-susceptibility (S) 913 sel General selectivity 669 e-sel Ecological selectivity 594 r-doc Documented pesticide resistance (R) 561 fctx Food chain toxicity 350 ec Ecological effects 339 mu Mode of pesticide uptake 227 r-crss Cross resistance (R) to pesticides 128 tb Toxicology or biochemistry (S) 125 rm-gi (R) management-genetic improvement 104 r-tg (R)-Toxicology or genetics 63 rm-i (R) management-introduced strains 63 md-r Methods development-resistance (R) 29 rm-t (R) management-theory 9 rm-e (R) management-endemic strains 3 Table 3. Distribution of SELCTV database records of pesticide impact on arthropod natural enmeis grouped by order. Order No. Records % Total Records Hymenoptera Coleoptera Acari Hemiptera Neuroptera Diptera Aranae Thysanoptera Dermaptera Leptidoptera 9.07 Scorpiones 3.02 Orthoptera 3.02 Odonata 3.02

33 21 By Natural Enemy Attributes Five levels of natural enemy identification were built into SELCTV record structure (Table 1). First, natural enemies were classified by type, i.e. predators or parasitoids. Predators accounted for 8993 records or just over 71% of SELCTV. Parasitoids made up 3591 records or approximately 29%. In 4 records, natural enemies were designated as "both" on a continuum between parasitoid and predator. Taxonomic characterization was initiated at the most general level. SELCTV pesticide impact data spanned thirteen orders of arthropod natural enemies (Table 3); 10 were in the class Insecta and 3 in the Arachnida. Hymenoptera was the most common order, accounting for over 27% of records. Coleoptera made up nearly 22% of SELCTV by record, followed by predaceous mites (Acari 19.2%), Hemiptera (14.1%) and the Neuroptera (8.7%). Natural enemies in SELCTV included some of the most important known biological control agents (Huffaker and Messenger 1976, Ridgway and Vinson 1977, Hoy and Herzog 1985, Belle and Sabelis 1986). Pesticide side effects in SELCTV spanned 88 families of arthropods. Of the 20 most numerous families three predators and one parasitoid comprised those most commonly tested for pesticide impact (i.e., predators include the phytoseiids, coccinellids and chrysopids). Thirteen of the top 20 families represented in SELCTV were generalist or mite predators. The Braconidae, Aphelinidae, Ichneumonidae and Trichogrammatidae, respectively, were the most commonly studied parasitoid families. Family composition within each order varied considerably. Hymenoptera was represented by 21 families in SELCTV, Coleoptera by 14, and Hemiptera and Araneae by 12 each. Data were extracted from the literature for 8 families of Acari and 7 of Diptera. The remaining 7

34 cl 1500 m cz) L.) U.. O cr 1000 w : N& Oda Oel Oda y:024 Ne Oda Nao y!" et1 oda ea ad' ack' e a, -o cf tcp a Nam da clo% (%1 (%' 0 SC ecel N.- clx.p'` 'f 0 (45 OP 9 Va ad* 'c43. CPC Cir Pi IC3 c>./ndqt, 1,0 Caer-r V.PC:S Vi x(4t. to t-9 No`c4 \ FAMILY Figure 3. Distribution of SELCTV database records of pesticide impact on arthropod natural enemies for the 20 most commonly tested families.

35 23 orders contained 3 or fewer families. These proportions become important when generalizing about susceptibility among natural enemies. For example, over 90% of acarine records detailed pesticide impact on the Phytoseiidae. The remaining 7 families made up only 9.6% of acarine records. Similarly, data for Coleoptera were 80% composed of the family Coccinellidae. In contrast, none of the 21 hymenopteran families accounted for more than about 30% of this order. SELCTV contained 342 unique genera (GENUS). Amblyseius made up 8% of database records (Table 4). Again, the 5 most frequently studied natural enemies, which accounted for nearly 30% of records, were either generalist or mite predators. The latter species have been documented as important biological control agents on a wide variety of high value crops (Croft and McGroarty 1977, Hoy et al. 1979, Berendt 1980, Penman and Chapman 1980, Cranham and Solomon 1981, Helle and Sabelis 1986). Trichogramma, Encarsia, Apanteles, Opius and Aphytis were the most commonly studied parasitoids. In comparison, these 5 parasitoid genera comprised less than 10% of SELCTV records. The 5 most commonly studied species were generalist or mite predators, led by Chrysopa carnea (Table 5). of the ranked top 20 species were either predatory phytoseiids or coccinellids. Encarsia formosa, 6th ranked by species, was the most prevalent parasitoid in SELCTV. E. formosa has been managed in greenhouse IPM programs, and has been extensively tested for susceptibility to relevant pesticides (e.g. Beglyarov and Maslienko 1978, Zseller and Budai 1982, Hassan et al. 1983, Hoogcarspel and Jobsen 1984). Only 6 of the 22 most common species were parasitoids, all of the order Hymenoptera. remainder were generalist or mite predators. The At this level of classification, a spider (Lycosa pseudoannulata) Ten

36 24 Table 4. Distribution of SELCTV database records of pesticide impact on arthropod natural enemies for the 20 most commonly tested genera. Genus Family No.Records Amblyseius Phytoseiidae 1009 Chrysopa Chrysopidae 986 Typhlodromus Phytoseiidae 629 Coccinella Coccinellidae 571 Hippodamia Coccinellidae 441 Trichogramma Trichogrammatidae 398 Stethorus Coccinellidae 379 Orius Anthocoridae 235 Encarsia Encyrtidae 233 Geocoris Lygaeidae 232 Apanteles Braconidae 222 Nabis Nabidae 212 Opius Braconidae 183 Aphytis Aphelinidae 179 Anthocoris Anthocoridae 174 Deraeocoris Miridae 169 Campoletis Ichneumonidae 147 Leptomastix Encyrtidae 139 Phygadeuon Ichneumonidae 124 Syrphus Syrphidae 117

37 25 Table 5. Distribution of SELCTV database records of pesticide impact on arthropod natural enemies for the 22 most commonly tested species. Species Family No.Records Chrysopa carnea Amblysieus fallacis Coccinella septempunctata Hippodamia convergens Phytoseiulus persimilis Typhlodromus occidentalis Encarsia formosa Typhlodromus pyri Trichogramma cacoeciae Chrysopa oculata Stethorus punctum Metasyrphus corollae Phygadeuon trichops Leptomastix dactylopii Cryptolaemus montrouzieri Coleomegilla maculata Hyaliodes harti Amblyseius hibisci Trichogramma evanescens Opius concolor Lycosa pseudoannulata Anthocoris nemorum Chrysopidae Phytoseiidae Coccinellidae Coccinellidae Phytoseiidae Phytoseiidae Aphelinidae Phytoseiidae Trichogrammatidae Chrysopidae Coccinellidae Syrphidae Ichneumonidae Encyrtidae Coccinellidae Coccinellidae Miridae Phytoseiidae Trichogrammatidae Braconidae Lycosidae Anthocoridae

38 26 made the list of prevalent species. In laboratory assessments of pesticide side effects where greater precision was possible, sex of the natural enemy was occasionally specified. Where referred to in the literature, sex was catalogued in NE:ATTRIBU (Natural enemy attributes). Sex is unspecified in over 90% of the records in SELCTV. In 8% of records, the test organisms were female. Slightly less than 2% of records document pesticide side effects on male natural enemies. Source of tested natural enemies was also entered in NE:ATTRIBU. Test organisms were either field collected (70%) reared (25%). or lab Susceptibility status (in NE:ATTRIBU) was assigned based on apparent pesticide toxicity to the natural enemy. The status was assigned as susceptible when natural enemy mortality was greater than 10%, accounting for about 71% of the database. Approximately 26% of records in SELCTV demonstrated tolerance, wherein natural enemies sustained only 10% mortality or less. Natural enemies which exhibited a 5-fold or greater survival ratio when exposed to a pesticide as compared to a known susceptible strain were termed resistant. Resistant natural enemies accounted for less than 3% of SELCTV or 347 records. Some level of resistance has been shown for selected species from the Aphelinidae, Braconidae, Carabidae, Cecidomyiidae, Coccinellidae, Encyrtidae, Lygaeidae, Phytoseiidae and Trichogrammatidae. However, some of these cases might have been more appropriately considered tolerance. Individual life stage (STAGE) was specified in over 65% of records in SELCTV (Fig. 4). Egg and pupal stages accounted for about 2% of records each. Susceptibility of immatures (larvae and nymphs) was documented in 1832 records (14.5%). Adults were most commonly tested at 46.7% of SELCTV records. In 32% of records, lifestage was